A system and method for searching images by images for multi-task bayonet vehicles

Abstract

The invention relates to a system and method for searching images by images for multi-task bayonet vehicles. The invention uses a deep neural network to establish a multi-task positioning and multi-task feature extraction network, based on an improved edge box detection technology and a cascaded loss function method Train the positioning network, respectively locate and feature detect the three parts of the vehicle in the bayonet vehicle image, the annual inspection mark and the lamp, combine the global and local features, and use the loss function of softmax loss and triple loss function to train the network , the final weighted combination of the local feature vector and the global feature vector of the last fully connected layer of the neural network is used as the vehicle feature for retrieval, and the improved k-means algorithm is used for retrieval to find out the K class, and then the SVM is used to form a hash function for Han Clear coding improves retrieval speed and saves storage space.

Description

technical field [0001] The invention relates to the field of intelligent transportation, in particular to a system and method for searching images by images for multi-task bayonet vehicles. Background technique [0002] With the development of society, intelligent traffic monitoring in the field of intelligent transportation is a very important development direction at present. At present, a large number of electronic police and checkpoint systems have been deployed on urban roads in my country. These systems can capture high-definition pictures of vehicles in real time, and identify and analyze the license plate number, as well as some vehicle model information (such as vehicle size, color, etc.). However, in the currently used bayonet monitoring system, the license plate number recognition still has a misrecognition and missed recognition rate of about 10%. More importantly, it will not be possible to identify illegal vehicles with license plates or deliberately concealed...

AI Technical Summary

Problems solved by technology

[0005] "Deep Learning-Based Vehicle Vehicle Identification Model Construction Method and Vehicle Vehicle Identification Method", application number 201610962720.5, using deep learning for vehicle vehicle identification, which has not reached the level of fine-grained vehicle retrieval and is not accurate enough

Images